bpf: Fix narrow load on a bpf_sock returned from sk_lookup()
Subsystems (4)
PATCH 1/1 bpf: Fix narrow load on a bpf_sock returned from sk_lookup()
By adding this test to test_verifier:
{
"reference tracking: access sk->src_ip4 (narrow load)",
.insns = {
BPF_SK_LOOKUP,
BPF_MOV64_REG(BPF_REG_6, BPF_REG_0),
BPF_JMP_IMM(BPF_JEQ, BPF_REG_0, 0, 3),
BPF_LDX_MEM(BPF_H, BPF_REG_2, BPF_REG_0, offsetof(struct bpf_sock, src_ip4) + 2),
BPF_MOV64_REG(BPF_REG_1, BPF_REG_6),
BPF_EMIT_CALL(BPF_FUNC_sk_release),
BPF_EXIT_INSN(),
},
.prog_type = BPF_PROG_TYPE_SCHED_CLS,
.result = ACCEPT,
},The above test loads 2 bytes from sk->src_ip4 where sk is obtained by bpf_sk_lookup_tcp().
It hits an internal verifier error from convert_ctx_accesses():
[root@arch-fb-vm1 bpf]# ./test_verifier 665 665
Failed to load prog 'Invalid argument'!
0: (b7) r2 = 0
1: (63) *(u32 *)(r10 -8) = r2
2: (7b) *(u64 *)(r10 -16) = r2
3: (7b) *(u64 *)(r10 -24) = r2
4: (7b) *(u64 *)(r10 -32) = r2
5: (7b) *(u64 *)(r10 -40) = r2
6: (7b) *(u64 *)(r10 -48) = r2
7: (bf) r2 = r10
8: (07) r2 += -48
9: (b7) r3 = 36
10: (b7) r4 = 0
11: (b7) r5 = 0
12: (85) call bpf_sk_lookup_tcp#84
13: (bf) r6 = r0
14: (15) if r0 == 0x0 goto pc+3
R0=sock(id=1,off=0,imm=0) R6=sock(id=1,off=0,imm=0) R10=fp0,call_-1 fp-8=????0000 fp-16=0000mmmm fp-24=mmmmmmmm fp-32=mmmmmmmm fp-40=mmmmmmmm fp-48=mmmmmmmm refs=1
15: (69) r2 = *(u16 *)(r0 +26)
16: (bf) r1 = r6
17: (85) call bpf_sk_release#86
18: (95) exit
from 14 to 18: safe
processed 20 insns (limit 131072), stack depth 48
bpf verifier is misconfigured
Summary: 0 PASSED, 0 SKIPPED, 1 FAILED
The bpf_sock_is_valid_access() is expecting src_ip4 can be narrowly loaded (meaning load any 1 or 2 bytes of the src_ip4) by marking info->ctx_field_size. However, this marked ctx_field_size is not used. This patch fixes it.
Due to the recent refactoring in test_verifier, this new test will be added to the bpf-next branch (together with the bpf_tcp_sock patchset) to avoid merge conflict.
--- kernel/bpf/verifier.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-)
kernel/bpf/verifier.c +7 -4
@@ -1617,12 +1617,13 @@ static int check_flow_keys_access(struct bpf_verifier_env *env, int off,return 0 ; } -static int check_sock_access (struct bpf_verifier_env *env,u32 regno,int off, -int size,enum bpf_access_type t) +static int check_sock_access (struct bpf_verifier_env *env,int insn_idx, +u32 regno,int off,int size, +enum bpf_access_type t) {struct bpf_reg_state *regs =cur_regs (env);struct bpf_reg_state *reg = ®s[regno]; -struct bpf_insn_access_aux info; +struct bpf_insn_access_aux info = {};if (reg->smin_value <0 ) { verbose(env, "R%d min value is negative, either use unsigned index or do a if (index >=0) check.\n",
@@ -1636,6 +1637,8 @@ static int check_sock_access(struct bpf_verifier_env *env, u32 regno, int off,return -EACCES ; } + env->insn_aux_data[insn_idx].ctx_field_size = info.ctx_field_size; +return 0 ; }
@@ -2032,7 +2035,7 @@ static int check_mem_access(struct bpf_verifier_env *env, int insn_idx, u32 regnverbose (env,"cannot write into socket\n" );return -EACCES ; } - err =check_sock_access (env, regno, off, size, t); + err =check_sock_access (env, insn_idx, regno, off, size, t);if (!err && value_regno >=0 )mark_reg_unknown (env, regs, value_regno); }else {
Signature · 1 line
2.17.1
Quoted 59 lines · Martin KaFai Lau
By adding this test to
test_verifier:{ "reference tracking: access sk->src_ip4 (narrow load)", .insns = { BPF_SK_LOOKUP, BPF_MOV64_REG(BPF_REG_6, BPF_REG_0), BPF_JMP_IMM(BPF_JEQ, BPF_REG_0, 0, 3), BPF_LDX_MEM(BPF_H, BPF_REG_2, BPF_REG_0, offsetof(struct bpf_sock, src_ip4) + 2), BPF_MOV64_REG(BPF_REG_1, BPF_REG_6), BPF_EMIT_CALL(BPF_FUNC_sk_release), BPF_EXIT_INSN(), }, .prog_type = BPF_PROG_TYPE_SCHED_CLS, .result = ACCEPT, },The above test loads 2 bytes from
sk->src_ip4where sk is obtained bybpf_sk_lookup_tcp().It hits an internal verifier error from
convert_ctx_accesses():
[root@arch-fb-vm1 bpf]# ./test_verifier 665 665
Failed to load prog 'Invalid argument'!
0: (b7) r2 = 0
1: (63) *(u32 *)(r10 -8) = r2
2: (7b) *(u64 *)(r10 -16) = r2
3: (7b) *(u64 *)(r10 -24) = r2
4: (7b) *(u64 *)(r10 -32) = r2
5: (7b) *(u64 *)(r10 -40) = r2
6: (7b) *(u64 *)(r10 -48) = r2
7: (bf) r2 = r10
8: (07) r2 += -48
9: (b7) r3 = 36
10: (b7) r4 = 0
11: (b7) r5 = 0
12: (85) call bpf_sk_lookup_tcp#84
13: (bf) r6 = r0
14: (15) if r0 ==0x0goto pc+3
R0=sock(id=1,off=0,imm=0) R6=sock(id=1,off=0,imm=0) R10=fp0,call_-1 fp-8=????0000 fp-16=0000mmmm fp-24=mmmmmmmm fp-32=mmmmmmmm fp-40=mmmmmmmm fp-48=mmmmmmmm refs=1
15: (69) r2 = *(u16 *)(r0 +26)
16: (bf) r1 = r6
17: (85) call bpf_sk_release#86
18: (95) exitfrom 14 to 18: safe
processed 20 insns (limit 131072), stack depth 48
bpf verifier is misconfigured
Summary: 0 PASSED, 0 SKIPPED, 1 FAILEDThe
bpf_sock_is_valid_access()is expectingsrc_ip4can be narrowly loaded (meaning load any 1 or 2 bytes of thesrc_ip4) by markinginfo->ctx_field_size. However, this markedctx_field_sizeis not used. This patch fixes it.Due to the recent refactoring in
test_verifier, this new test will be added to the bpf-next branch (together with thebpf_tcp_sockpatchset) to avoid merge conflict.
Quoted 4 lines
Fixes c64b7983288e ("bpf: Add PTR_TO_SOCKET verifier type")Cc Joe Stringer <[email protected]>Signed-off-by Martin KaFai Lau <[email protected]>---
Nice find, thanks.
Alexei Starovoitov
AppliedQuoted 53 lines · Martin KaFai Lau
By adding this test to
test_verifier:{ "reference tracking: access sk->src_ip4 (narrow load)", .insns = { BPF_SK_LOOKUP, BPF_MOV64_REG(BPF_REG_6, BPF_REG_0), BPF_JMP_IMM(BPF_JEQ, BPF_REG_0, 0, 3), BPF_LDX_MEM(BPF_H, BPF_REG_2, BPF_REG_0, offsetof(struct bpf_sock, src_ip4) + 2), BPF_MOV64_REG(BPF_REG_1, BPF_REG_6), BPF_EMIT_CALL(BPF_FUNC_sk_release), BPF_EXIT_INSN(), }, .prog_type = BPF_PROG_TYPE_SCHED_CLS, .result = ACCEPT, },The above test loads 2 bytes from
sk->src_ip4where sk is obtained bybpf_sk_lookup_tcp().It hits an internal verifier error from
convert_ctx_accesses():
[root@arch-fb-vm1 bpf]# ./test_verifier 665 665
Failed to load prog 'Invalid argument'!
0: (b7) r2 = 0
1: (63) *(u32 *)(r10 -8) = r2
2: (7b) *(u64 *)(r10 -16) = r2
3: (7b) *(u64 *)(r10 -24) = r2
4: (7b) *(u64 *)(r10 -32) = r2
5: (7b) *(u64 *)(r10 -40) = r2
6: (7b) *(u64 *)(r10 -48) = r2
7: (bf) r2 = r10
8: (07) r2 += -48
9: (b7) r3 = 36
10: (b7) r4 = 0
11: (b7) r5 = 0
12: (85) call bpf_sk_lookup_tcp#84
13: (bf) r6 = r0
14: (15) if r0 ==0x0goto pc+3
R0=sock(id=1,off=0,imm=0) R6=sock(id=1,off=0,imm=0) R10=fp0,call_-1 fp-8=????0000 fp-16=0000mmmm fp-24=mmmmmmmm fp-32=mmmmmmmm fp-40=mmmmmmmm fp-48=mmmmmmmm refs=1
15: (69) r2 = *(u16 *)(r0 +26)
16: (bf) r1 = r6
17: (85) call bpf_sk_release#86
18: (95) exitfrom 14 to 18: safe
processed 20 insns (limit 131072), stack depth 48
bpf verifier is misconfigured
Summary: 0 PASSED, 0 SKIPPED, 1 FAILEDThe
bpf_sock_is_valid_access()is expectingsrc_ip4can be narrowly loaded (meaning load any 1 or 2 bytes of thesrc_ip4) by markinginfo->ctx_field_size. However, this markedctx_field_sizeis not used. This patch fixes it.
Quoted 8 lines
Due to the recent refactoring in
test_verifier, this new test will be added to the bpf-next branch (together with thebpf_tcp_sockpatchset) to avoid merge conflict.Fixes c64b7983288e ("bpf: Add PTR_TO_SOCKET verifier type")Cc Joe Stringer <[email protected]>Signed-off-by Martin KaFai Lau <[email protected]>
Applied to bpf tree.
Martin, if your is_fullsock work depends on it, I can apply the fix to bpf-next as well. Just let me know.
Quoted 59 lines · Alexei Starovoitov
Quoted 58 lines · Martin KaFai Lau
By adding this test to
test_verifier:{ "reference tracking: access sk->src_ip4 (narrow load)", .insns = { BPF_SK_LOOKUP, BPF_MOV64_REG(BPF_REG_6, BPF_REG_0), BPF_JMP_IMM(BPF_JEQ, BPF_REG_0, 0, 3), BPF_LDX_MEM(BPF_H, BPF_REG_2, BPF_REG_0, offsetof(struct bpf_sock, src_ip4) + 2), BPF_MOV64_REG(BPF_REG_1, BPF_REG_6), BPF_EMIT_CALL(BPF_FUNC_sk_release), BPF_EXIT_INSN(), }, .prog_type = BPF_PROG_TYPE_SCHED_CLS, .result = ACCEPT, },The above test loads 2 bytes from
sk->src_ip4where sk is obtained bybpf_sk_lookup_tcp().It hits an internal verifier error from
convert_ctx_accesses():
[root@arch-fb-vm1 bpf]# ./test_verifier 665 665
Failed to load prog 'Invalid argument'!
0: (b7) r2 = 0
1: (63) *(u32 *)(r10 -8) = r2
2: (7b) *(u64 *)(r10 -16) = r2
3: (7b) *(u64 *)(r10 -24) = r2
4: (7b) *(u64 *)(r10 -32) = r2
5: (7b) *(u64 *)(r10 -40) = r2
6: (7b) *(u64 *)(r10 -48) = r2
7: (bf) r2 = r10
8: (07) r2 += -48
9: (b7) r3 = 36
10: (b7) r4 = 0
11: (b7) r5 = 0
12: (85) call bpf_sk_lookup_tcp#84
13: (bf) r6 = r0
14: (15) if r0 ==0x0goto pc+3
R0=sock(id=1,off=0,imm=0) R6=sock(id=1,off=0,imm=0) R10=fp0,call_-1 fp-8=????0000 fp-16=0000mmmm fp-24=mmmmmmmm fp-32=mmmmmmmm fp-40=mmmmmmmm fp-48=mmmmmmmm refs=1
15: (69) r2 = *(u16 *)(r0 +26)
16: (bf) r1 = r6
17: (85) call bpf_sk_release#86
18: (95) exitfrom 14 to 18: safe
processed 20 insns (limit 131072), stack depth 48
bpf verifier is misconfigured
Summary: 0 PASSED, 0 SKIPPED, 1 FAILEDThe
bpf_sock_is_valid_access()is expectingsrc_ip4can be narrowly loaded (meaning load any 1 or 2 bytes of thesrc_ip4) by markinginfo->ctx_field_size. However, this markedctx_field_sizeis not used. This patch fixes it.Due to the recent refactoring in
test_verifier, this new test will be added to the bpf-next branch (together with thebpf_tcp_sockpatchset) to avoid merge conflict.
Quoted 5 lines
Quoted 3 lines
Fixes c64b7983288e ("bpf: Add PTR_TO_SOCKET verifier type")Cc Joe Stringer <[email protected]>Signed-off-by Martin KaFai Lau <[email protected]>Applied to bpf tree.
Thanks!
Quoted 3 lines
Martin, if your
is_fullsockwork depends on it, I can apply the fix to bpf-next as well. Just let me know.
Yes, the is_fullsock work depends on it.
I should have mentioned it in this commit log.
Quoted 4 lines · Martin Lau
Quoted 4 lines
Quoted 3 lines
Fixes c64b7983288e ("bpf: Add PTR_TO_SOCKET verifier type")Cc Joe Stringer <[email protected]>Signed-off-by Martin KaFai Lau <[email protected]>
Quoted 8 lines
Quoted 1 line
Applied to bpf tree.
Thanks!
Quoted 3 lines
Martin, if your
is_fullsockwork depends on it, I can apply the fix to bpf-next as well. Just let me know.Yes, the
is_fullsockwork depends on it.
I should have mentioned it in this commit log.
Ok. I've pushed it to bpf-next as well.
Last time we discusses this scenario at netconf and agreed that git should do the right thing, since commit is the same.
I think this is a case where I think it makes sense to give it a shot. If we get issues during pulls/merges it will be a lesson to avoid such things in the future, but if we don't try it we won't know. So applied.